Combined trust region methods for nonlinear least squares

beta

Combined trust region methods for nonlinear least squares

L. Luksan

1996

Citations

Journal

Kybernetika

Semantic Scholar

Key Takeaway

Key Takeaway: This paper presents three efficient algorithms for nonlinear least squares problems, improving standard trust region techniques and demonstrating their effectiveness through extensive numerical experiments.

Abstract

Trust region realizations of the Gauss Newton method are commonly used for obtaining solution of nonlinear least squares problems We propose three e cient algorithms which improve standard trust region techniques multiple dog leg strat egy for dense problems and two combined conjugate gradient Lanczos strategies for sparse problems E ciency of these methods is demonstrated by extensive numerical experiments

Text copied to clipboard

copied to clipboard